Network Attack Detection System Research And Design Based On Embeded System

With the coming of many kinds of network invading and attack, there are also many product made for anti-network invading and attack appear in market. Each network security product focus on different aspect aiming at different way of invading. The most widly known network attack is DDOS which originate from the defect of underlying network protocol,and there is no technology to deal with the problem until now. so many security manufacturers and research center take measures of detection and prevention such as intrusion detection system which used Neural Network technology and has wide range of application on large-scale system, it is a more mature technology. Owing to embeded system’s low power consumption, performant focused and high mobility, the combination of Network Security and Embedded System would become new direction of technology development. In this article, we use SNMP Protocol to access to network performance data, combined Embedded System and Neural Network to achieve Situational awareness about Network Security in small-scale network environment.SNMP protocol (Simple Network Managerment Protocol),was designed to remoted manage device and PC for network administrator who can access to data and category of device’s operation status,then have a decision-making. in addition network administrator can take the initiative to set SNMP’s parameter, this measure make device and PC to respond flexibly in different environments. neural network originated in Biological Neurology and math,it is a kind of way of Artificial Intelligence and Machine Learning through simulation of biological neural process.in the beginning neural network just as a theoretical research the widely used in the engineering field,and now it has become the forefront of technology development and important area of research of industry.at first simulate biological neuron’s reaction process to external stimulation with caiculation, then set up a network with multiple neurons to work together to achieve complicated function. the article based above-mention theory,combine the three,SNMP,neural network,embeded system to one unity.The article’s achievement:(1implemented key modules obtained using SNMP performance data object as the data acquisition network security detection system, studies the SNMP protocol data acquisition technology and network security status Perception.(2)find out a SNMP-based performance data to the network status classification, and anomaly detection methods, Get operating data using SNMP protocol,through Multiple processing,Explore the trends of the data behind the state of the device.(3) designed BP neural network structure which can work in the low-performance embedded environment,and comfirmed the whole structure of neural network with the method that the radial basis function neural network hidden layer node used.designed and implemented network security situational awareness system using library function from open source project named net-snmp. about implication for research:(1)because General network security situational awareness system is based on a large desktop system or server,is a very complex system with very complicated control and communication,but this article will focus on particular small scale and weak Environment,remove unnecessary complex models and simplify complex issues.(2)based on the work of predecessors combined SNMP-based data acquisition technology and neural network to design Perceptual system based on the state of the statistical model conversion,combined SNMP protocol network management technology with neural network and embedded system..